Microsoft to scale in-house AI chips to multi-gigawatt deployments from 2027
Microsoft plans to expand its Azure Maia AI chip family to multi-gigawatt deployments starting 2027, aiming to make it the primary system for AI models. The company introduced Maia 300, expected to deliver 40% more tokens per watt than leading GPUs. Maia 200, already deployed, offers 1.7x more tokens per watt in Microsoft Teams and Copilot. The chips are designed for efficiency and demanding AI workloads, with Maia 300 expected to have seven times the token throughput of Maia 200.
How this was made

The 30-second read
Why it matters
The announcement may lift Microsoft shares on expectations of higher Azure margins and reduced reliance on external GPU pricing.
Market read
A strategic shift to in‑house AI chips could enhance Microsoft’s competitive position in cloud AI services, affecting both the stock and the broader AI hardware market.
What to watch
Supply‑chain constraints at TSMC and the need for ecosystem software support could temper near‑term impact.
Background
Microsoft's AI strategy has relied on third‑party GPUs; the Maia family marks a move to custom silicon for cost and performance advantages.
Ticker impact
Microsoft announced its in‑house Azure Maia AI chip family will scale to multi‑gigawatt deployments starting in 2027, introducing the Maia 300 with 40% more tokens per watt.
likely upward pressure as investors price in higher AI services profitability
First‑time disclosure of a large‑scale, energy‑efficient AI chip strategy that could differentiate Azure and capture more AI spend.
Market effects
Strengthens the AI‑hardware segment and may pressure competitors like Nvidia and AMD to accelerate their own silicon roadmaps.
Highlights U.S. leadership in AI infrastructure, potentially attracting global enterprise spend to Azure.
Signals a shift toward proprietary AI chips, influencing broader tech‑sector valuations.
Counterpoint
Execution risk and high capital spend could delay benefits, and competitors may retain market share if Microsoft's chips underperform.
Key entities
- ExecutiveSatya Nadella
CEO of Microsoft, announced the AI chip scaling plan.
- ExecutiveScott Guthrie
Executive Vice President for Cloud and AI, detailed production rollout.


